A Method for Customer Review Sentiment Analysis Based on Hybrid Classification Models to Automate Feedback in IT Project Management
Measuring and Computing Devices in Technological Processes, No. 3, pp. 487-496
Editorial summary
Customer reviews are classified by combining support-vector probabilities with a neural network and two-word text features. On the study’s balanced dataset, accuracy was 89.0% for the hybrid, 88.9% for the neural network and 86.2% for the support-vector classifier. The measured gain over the neural network alone was small.
Bibliographic reference
Use the publisher’s record for the citation format required by your journal or organisation.
Bohdan Zahurskyi, Eduard Manziuk, Tetiana Skrypnyk, Pavlo Radiuk. "Метод аналізу тональності відгуків клієнтів на основі гібридних моделей класифікації для автоматизації зворотного зв’язку в управлінні ІТ-проєктами". Measuring and Computing Devices in Technological Processes, No. 3, pp. 487-496, 2026. https://doi.org/10.31891/2219-9365-2026-87-56
Test a method against your business task.
Describe the decision your system needs to support and the data you have. We assess whether a method fits, define a prototype and agree how to measure its quality, response time and operating cost.